The Biggest Obstacle to Personalization Is the Creative Element

In a world where everyone is exposed to constant marketing through every conceivable media channel every day, messages that are not relevant to the target will be utterly ignored. And don’t blame the consumers for it, either. You, as a consumer, are trained to ignore irrelevant messages, as well.

In this consumer-centric environment, personalization is something all marketers must practice constantly, not only to increase the level of customer engagement, but also to not be ignored completely. And if your messages keep getting ignored, decreasing click-through rate isn’t just some annoying KPI that doesn’t look good in front of your boss, it may be an existential threat to your organization.

Unfortunately, personalization isn’t easy, simple, or cheap. There are many elements that must work harmoniously, so that each target sees something that is uniquely relevant to “her.”

4 Elements of Personalization

First, you need data about the target. What is she about, and what does she look like? That may require data from all kinds of sources — be they online or offline transactions, browsing history, store visits, reactions to previous campaigns (requiring both campaign and response history data), call-center logs, third-party demographic data, etc. Putting them all in one place, and rearranging them to create coveted Customer-360 View is often the first hurdle. But that is just the beginning. Without customer-centric data, there is no personalization — unless you count on your guesswork.

Then you need to make sense out of collected data. We often call such work analytics, which includes segmentation (or clustering), modeling, personas development (a series of affinity models), etc. Many marketers consider this to be the highest hurdle, as it requires different types of talents. Data scientists tend to think that the modeling work is the pinnacle of personalization, and they may not be wrong. But is it enough? So, what if they have 40 personas meticulously built by top-notch statisticians? How would you use them to differentiate messages for “each” target?

That leads to the third and forth elements in personalization, which are “Display Capability” and “Content and Creative.” Basically, you need to be able to show different creatives to different targets. If you are uniformly displaying the same content to everyone, what is the point in all this, no matter how many personas or affinity models you built?

Display capability is a technical hurdle. And you can procure technologies to overcome it, whether the challenge is dynamic web content, or personalized email delivery. You have to align pieces of technologies to make it happen. If Person A shows up on your website, and her affinity score is higher for “Luxury Travel” category in comparison to “Family Oriented Activities,” you should be able to show a picture of luxury cruise ship sailing in the Caribbean sunset, not necessarily a picture of happy children surrounded by cartoon characters.

As you can see, I am actually mixing three elements in this one example. I am assuming you built a series of personas (or affinity models). Your website should be dynamic so that such models can trigger different experiences for different visitors. Then of course, I am assuming you have ample amount of marketing creatives to differentiate messages. Display technology is a prerequisite in all this. If you don’t have it, go get it.

Your Persona Menu

Building a Customer-360 View is a customer-centric activity, but creating a persona menu is a selfish activity. What do you want to sell? And what kind of person would be interested in such products or services?

If you are selling fashion items, personas such as “Fashionista” or “Trend Setter” would be helpful. If you are pushing cutting-edge products, an “Early Adopter” persona would be necessary. If you are selling various types of insurance or security-related products, you will benefit from personas such as “Security Conscious.”

The important point here is that you should create persona menu based on your product and marketing roadmap. Be imaginative and creative. What kind of persona would be interested in your services? Once the goal is set, we need some samples of people who actually displayed such tendencies or behaviors. If you are building a persona called “Luxury Travel,” gather samples of people who actually have been on a luxury cruise ship or checked into luxury hotels (of course you have to define what constitutes “luxury”). Modelers do the rest.

Now, here is the reason why setting up a proper persona menu is so important. Not only will we define the target audience with it, but also categorize your marketing contents and digital assets with personas.

The most basic usage of any model is to go after high score individuals in a given category. You want to send messages to fashion-oriented people? Just select high score individuals using the Fashionista model.

But personalization is a little more complex that that. Let’s just say this one individual showed up at your website (or your store for that matter). You may have less than one second to show something that “she” would be interested in. Pull up all persona scores for that person, and see in which categories she scores high (let’s say over 7 out of a maximum score of 9). Going back to the previous example, if the target has score of 8 in Luxury Travel, and 4 in Family-oriented Activity, pull out the content for the former.

The Creative Element

Now, why is this article titled “The Biggest Obstacle to Personalization Is the Creative Element”? Because, I often see either lack of enough creative materials or lack of proper content library is the roadblock. And it really breaks my heart. With all the dynamic display capabilities and a series of models and personas, it would be a real shame if everyone gets to see the same damn picture.

I’ve seen sad and weird cases where marketers balk at the idea of personalization, as their creative agency is not flexible enough to create multiple versions of marketing materials. In this day and age, that is just a horrible excuse. What are they dealing with, some Mad Men agency people from the 1950s with cigarettes in their mouths and glasses of Scotch in their hands?

I’ve also seen other strange cases where proper personalization doesn’t happen – even with all good elements ready to be deployed – because departments don’t know how to communicate with one another. That is why someone should be in charge of all four elements of personalization.

How will the persona menu be created with grand marketing goals in mind? Who would procure actual data and build models? How will the resultant model/persona scores be shared throughout the organization and various systems, especially with the dynamic display technologies? How will the content library be tagged with all the relevant “persona” names (e.g., Tag “Luxury Travel” persona name to all digital assets related to “Luxury Cruise Ships”)?

Model scores (or personas) may function as a communication tool that binds different departments and constituents. Personalization is a team sport, and it is only as good as the weakest link. If you invested in building CDP solutions and analytics, go a little further and finish the work with the creative elements.

If you have a bunch of pictures stored in someone’s PC (or worse, some agency guy’s drawer), go build a digital content library. And while you’re at it, tag those digital assets with relevant persona names in your persona menu. Even automated personalization engines would appreciate your effort, and it will definitely pay off.

Models Are Built, But the Job Isn’t Done Yet

In my line of business – data and analytics consulting and coaching – I often recommend some modeling work when confronted with complex targeting challenges. Through this series, I’ve shared many reasons why modeling becomes a necessity in data-rich environments (refer to “Why Model?”).

The history of model-based targeting goes back to the 1960’s, but what is the number one reason to employ modeling techniques these days? We often have too much information, way beyond the cognitive and arithmetical capacities of our brains. Most of us mortals cannot effectively consider more than two or three variables at a time. Conversely, machines don’t have such limitations when it comes to recognizing patterns among countless data variables. Subsequent marketing automation is just an added bonus.

We operate under a basic assumption that model-based targeting (with deep data) should outperform some man-made rules (with a handful of information). At times, however, I get calls as campaign results prove otherwise. Sometimes campaign segments selected by models show worse response rates than randomly selected test groups do.

When such disappointing results happen, most decision makers casually say, “The model did not work.” That may be true, but more often than not, I find that something went wrong “before” or “after” the modeling process. (Refer to “Know What to Automate With Machine Learning”, where I list major steps concerning the “before” of model-based targeting).

If the model is developed in an “analytics-ready” environment where most input errors are eradicated, then here are some common mishaps in post-modeling stages to consider.

Mishap #1: The Model Is Applied to the Wrong Universe

Model algorithm is nothing but a mathematical expression between target and comparison universes. Yes, setting up the right target is the key for success in any modeling, but defining a proper comparison universe is equally important. And the comparison group must represent the campaign universe to which the resultant model is applied.

Sometimes such universes are defined by a series of pre-selection rules before the modeling even begins. For example, the campaign universes may be set by region (or business footprint), gender of the target, availability of email address or digital ID, income level, home ownership, etc. Once set, the rules must be enforced throughout the campaign execution.

What if the rules that define the modeling universe are even slightly different from the actual campaign universe? The project may be doomed from the get-go.

For example, do not expect that models developed within a well-established business footprint will be equally effective in relatively new prospecting areas. Such expansion calls for yet another set of models, as target prospects are indeed in a different world.

If there are multiple distinct segments in the customer base, we often develop separate models within each key segment. Don’t even think about applying a model developed in one specific segment to another, just because they may look similar on the surface. And if you do something like that, don’t blame the modeler later.

Mishap #2: The Model Is Used Outside Design Specification

Even in the same modeling universe, we may develop multiple types of models for different purposes. Some models may be designed to predict future lifetime value of customers, while others are to estimate campaign responsiveness. In this example, customer value and campaign responsiveness may actually be inversely related (e.g., potential high value customers less likely to be responsive to email campaigns).

If multiple response models are built for specific channels, do not use them interchangeably. Each model should be describing distinct channel behaviors, not just general responsiveness to given offers or products.

I’ve seen a case where a cruise ship company used an affinity model specifically designed for a seasonal European line for general purposes in the name of cost savings. The result? It would have been far more cost effective developing another model than having to deal with the fallout from ineffective campaigns. Modeling cost is often a small slice in the whole pie of campaign expenses. Don’t get stingy on analytics and call for help when in doubt.

Mishap #3: There Are Scoring Errors

Applying a model algorithm to a validation sample is relatively simple, as such samples are not really large. Now, try to apply the same algorithm to over 100 million potential targets. You may encounter all kinds of performance issues caused by the sheer volume of data.

Then there are more fundamental errors stemming from the database structure itself. What if the main database structure is different from that of the development sample? That type of discrepancy – which is very common – often leads to disasters.

Always check if anything is different between the development samples and the main database:

  • Database Structure: There are so many types of database platforms, and the way they store simple transaction data may be vastly different. In general, to rank individuals, each data record must be scored on an individual level, not transaction or event levels. It is strongly recommended that data consolidation, summarization, and variable creation be done in an analytics-friendly environment “before” any modeling begins. Structural consistency eliminates many potential errors.
  • Variable List/Names: When you have hundreds, or even thousands of variables in the database, there will be similar sounding names. I’ve seen many different variable names that may represent “Total Individual Dollar Amount Past 12-month,” for example. It is a common mistake to use a wrong data field in the scoring process.
  • Variable Values: Not all similar sounding variables have similar values in them. For example, ever-so-popular “Household Income” may include dollar values in thousand-dollar increments, or pre-coded value that looks like alphabets. What if someone changed the grouping definition of such binned variables? It would be a miracle if the model scores come out correctly.
  • Imputation Assumptions: There are many ways to treat missing values (refer to “Missing Data Can Be Meaningful”). Depending on how they were transformed and stored, even missing values can be predictable in models. If missing values are substituted with imputed values, it is absolutely important to maintain their consistency throughout the process. Mistreatment of missing values is often the main cause for scoring errors.

Mishap #4: Nature of Data Is Significantly Shifted

Data values change over time due to outside factors. For instance, if there is a major shift in the business model (e.g., business moving to a subscription model), or a significant change in data collection methods or vendors, consider that all the previous models are now rendered useless. Models should be predictors of customer behaviors, not reflections of changes in your business.

Mishap #5: Scores Are Tempered After-the-Fact

This one really breaks my heart, but it happens. I once saw a user in a major financial institution unilaterally change the ranges of model decile groups after observing significant fluctuations in model group counts. As you can imagine by now, uneven model group counts are indeed revealing serious inconsistencies caused by any of the factors that I mentioned thus far. You cannot tape over a major wound — just bite the bullet and commission a new model when you see uneven or inconsistent model decile counts.

Mishap #6: There Are Selection Errors

When campaign targets are selected based on model scores, the users must be fully aware of the nature of them. If the score is grouped into model groups 1 through 10, is the ideal target “1” or “10”?

I’ve seen cases where the campaign selection was completely off the mark, as someone sorted the raw score in an ascending order, not a descending order, pushing the worse prospects to the top. But I’ve also seen errors in documentation or judgement, as it can be really confusing to figure out which group is “better.”

I tend to put things in 0-9 scale when designing a series of personas or affinity models to avoid confusion. If score groups range from 0 to 9, the user is much less likely to assume that “zero” is the best score. Without a doubt, reversed score is far worse than not using the model at all.

Final Thoughts

After all, the model algorithm itself can be wrong, too. Not all modelers are equally competent, and machine-learning is only as good as the analyst who originally set it up. Of course, you must turn that stone when investigating bad results. But you should trace all pre- and post-modeling steps, as well. After years of such detective work, my bet is firmly on errors outside the modeling processes, unless the model validation smells fishy.

In any case, do not entirely give up on modeling just because you’ve had a few bad results. There are many things to be checked and tweaked, and model-based targeting is a long series of iterative adjustments. Be mindful that even a mediocre model is still better than someone’s gut feelings, if it is applied to campaigns properly.

Know What to Automate With Machine Learning

There are many posers in the data and analytics industry. Unfortunately, some of them occupy managerial positions, making critical decisions based on superficial knowledge and limited experiences. I’ve seen companies wasting loads of money and resources on projects with no substantial value — all because posers in high places bought into buzzwords or false promises. As if buzzwords have some magical power to get things done “auto-magically.”

I’ve written articles about how to identify posers and why buzzwords suck. But allow me to add a few more thoughts, as the phrase “Machine Learning” is rapidly gaining that magical power in many circles. You’d think that machines could read our minds and deliver results on their own. Sorry to break it to you, but even in the world of Star Trek, computers still wouldn’t understand illogical requests.

Beware of people who try to employ machine learning and no other technique. Generally, such people don’t even understand what they are trying to automate, only caring about the cost reduction part. But the price that others end up paying for such a bad decision could be far greater than any savings. The worst-case scenario is automating inadequate practices, which leads to wrong places really fast. How can anyone create a shortcut if he doesn’t know how to get to the destination in the first place, or worse, where the destination is supposed to be?

The goal of any data project should never be employing machine learning for the sake of it. After all, you wouldn’t respect a guitarist who can’t play a simple lick, just because he has a $5,000 custom guitar on his shoulder.

Then, what is the right way to approach this machine learning hype? First, you must recognize that there are multiple steps in predictive modeling. Allow me to illustrate some major steps and questions to ask:

  1. Planning: This critical step is often the most difficult one. What are you trying to achieve through data and analytics? Building the most eloquent model can’t be the sole purpose outside academia. Converting business goals into tangible solution sets is a project in itself. What kind of analytics should be employed? What would be the outcome? How will those model scores be applied to actual marketing campaigns? How will the results would be measured? Prescribing proper solutions to business challenges within the limitation of systems, toolsets, and the budget is one of the most coveted skill sets. And it has nothing to do with tools like machine learning, yet.
  2. Data Audit: Before we chart a long analytics journey, let’s put a horse before the cart, as data is the fuel for an engine called machine learning. I’ve seen too many cases where the cart is firmly mounted before the horse. What data are we going to use? From what sources? Do we have enough data to perform the task? How far in time do the datasets go back? Are they merged in one place? Are they in usable forms? Too many datasets are disconnected, unstructured, uncategorized, and unclean. Even for the machines.
  3. Data Transformation: Preparing available data for advanced analytics is also a project in itself. Be mindful that you don’t have to clean everything; just deal with the elements that are essential for required analytics to meet pre-determined business goals. At this stage, you may employ machine learning to categorize, group, or reformat data variables. But note that such modules are quite different from the ones for predictions.
  4. Target Definition: Setting up proper model targets is half-art/half-science. If the target is hung on a wrong spot, the resultant model will never render any value. For instance, if you are targeting so-called “High Value” customers, how would you express it in mathematical terms? It could be defined by any combinations of value, frequency, recency, and product categories. The targets are to be set after a long series of assumptions, profiling, and testing. No matter what modeling methodology eventually gets employed, you do NOT want targets to be unilaterally determined by a machine. Even with a simple navigator, which provides driving directions through machine-based algorithms, the user must provide the destination first. A machine cannot determine where you need to go (at least not yet).
  5. Universe Definition: In what universe will the resultant model be applied and used? Model comparison universe is as important as the target itself, as a model score is a mathematical expression of differences between two dichotomous universes (e.g., buyers vs. non-buyers). Even with the same target, switching the comparison universe would render completely different algorithms. On top of that, you may want to put extra filters by region, gender, customer type, user segment, etc. A machine may determine distinct sets of universes that require separate models, but don’t relinquish all controls to machines, either. Machine may not aware of where you would apply the model.
  6. Modeling: This statistical work is comprised of sub-steps such as variable selection, variable transformation, binning, outlier exclusion, algorithm creation, and validation, all in multiple iterations. It is indeed laborious work, and “some” parts may be done by the machines to save time. You may have heard of terms such as Deep Learning, Neural Net, logistic regression, stepwise regression, Random Forest, CHAID analysis, tree analysis, etc. Some are to be done by machines, and some by human analysts. All those techniques are basically to create algorithms. In any case, some human touch is inevitable regardless of employed methodology, as nothing should be released without continuous testing, validation, and tweaking. Don’t blindly subscribe to terms like “unsupervised learning.”
  7. Application: An algorithm may have been created in a test environment, but to be useful, the model score must be applied to the entire universe. Some toolsets provide “in-database-scoring”, which is great for automation. Let me remind you that most errors happen before or after the modeling step. Again, humans should not be out of the loop until everything becomes a routine, all the way to campaign execution and attribution.
  8. Maintenance: Models deteriorate and require scheduled reviews. Even self-perpetuating algorithms should be examined periodically, as business environments, data quality, and assumptions may take drastic turns. The auto-pilot switch shouldn’t stay on forever.

So, out of this outline for a simple target modeling (for 1:1 marketing applications), which parts do you think can fully be automated without any human intervention? I’d say some parts of data transformation, maybe all of modeling, and some application steps could go on the hands-free route.

The most critical step of all, of course, is the planning and goal-setting part. Humans must breathe their intention into any project. Once things are running smoothly, then sure, we can carve out the parts that can be automated in a step-wise fashion (i.e., never in one shot).

Now, would you still believe sales pitches that claim all your marketing dreams will come true if you just purchase some commercial machine-learning modules? Even if decent toolsets are tuned up properly, don’t forget that you are supposed to be the one who puts them in motion, just like self-driving cars.

Beware of One-Size-Fits-All Customer Data Solutions

In the data business, the ability to fine-tune database structure and toolsets to meet unique business requirements is key to success, not just flashy features and functionalities. Beware of technology providers who insist on a “one-size-fits-all” customer data solution.

In the data business, the ability to fine-tune database structure and toolsets to meet unique business requirements is key to success, not just flashy features and functionalities. Beware of technology providers who insist on a “one-size-fits-all” customer data solution, unless the price of entry is extremely low. Always check the tech provider’s exception management skills and their determination to finish the last mile. Too often, many just freeze at the thought of any customization.

The goal of any data project is to create monetary value out of available data. Whether it is about increasing revenue or reducing cost, data activities through various types of basic and advanced analytics must yield tangible results. Marketers are not doing all this data-related work to entertain geeks and nerds (no offense); no one is paying for data infrastructure, analytics toolsets, and most importantly, human cost to support some intellectual curiosity of a bunch of specialists.

Therefore, when it comes to evaluating any data play, the criteria that CEOs and CFOs bring to the table matter the most. Yes, I shared a long list of CDP evaluation criteria from the users’ and technical points of views last month, but let me emphasize that, like any business activity, data work is ultimately about the bottom line.

That means we have to maintain balance between the cost of doing business and usability of data assets. Unfortunately, these two important factors are inversely related. In other words, to make customer data more useful, one must put more time and money into it. Most datasets are unstructured, unrefined, uncategorized, and plain dirty. And the messiness level is not uniform.

Start With the Basics

Now, there are many commoditized toolsets out in the market to clean the data and weave them together to create a coveted Customer-360 view. In fact, if a service provider or a toolset isn’t even equipped to do the basic part, I suggest working with someone who can.

For example, a service provider must know the definition of dirty data. They may have to ask the client to gauge the tolerance level (for messy data), but basic parameters must be in place already.

What is a good email address, for instance? It should have all the proper components like @ signs and .com, .net, .org, etc. at the end. Permission flags must be attached properly. Primary and secondary email must be set by predetermined rules. They must be tagged properly if delivery fails, even once. The list goes on. I can think of similar sets of rules when it comes to name, address, company name, phone number, and other basic data fields.

Why are these important? Because it is not possible to create that Customer-360 view without properly cleaned and standardized Personally Identifiable Information (PII). And anyone who is in this game must be masters of that. The ability to clean basic information and matching seemingly unmatchable entities are just prerequisites in this game.

Even Basic Data Hygiene and Matching Routines Must Be Tweaked

Even with basic match routines, users must be able to dictate tightness and looseness of matching logics. If the goal of customer communication involves legal notifications (as for banking and investment industries), one should not merge any two entities just because they look similar. If the goal is mainly to maximize campaign effectiveness, one may merge similar looking entities using various “fuzzy” matching techniques, employing Soundex, nickname tables, and abbreviated or hashed match keys. If the database is filled with business entities for B2B marketing, then so-called commoditized merge rules become more complicated.

The first sign of trouble often becomes visible at this basic stage. Be aware of providers that insist on “one-size-fits-all” rules, in the name of some universal matching routine. There was no such thing even in the age of direct marketing (i.e., really old days). How are we going to go through complex omnichannel marketing environment with just a few hard-set rules that can’t be modified?

Simple matching logic only with name, address, and email becomes much more complex when you add new online and offline channels, as they all come with different types of match keys. Just in the offline world, the quality of customer names collected in physical stores vastly differs from that of self-entered information from a website along with shipping addresses. For example, I have seen countless invalid names like “Mickey Mouse,” “Asian Tourist,” or “No Name Provided.” Conversely, no one who wants to receive the merchandise at their address would create an entry “First Name: Asian” and “Last Name: Tourist.”

Sure, I’m providing simple examples to illustrate the fallacy of “one-size-fits-all” rules. But by definition, a CDP is an amalgamation of vastly different data sources, online and offline. Exceptions are the rules.

Dissecting Transaction Elements

Up to this point, we are still in the realm of “basic” stuff, which is mostly commoditized in the technology market. Now, let’s get into more challenging parts.

Once data weaving is done through PII fields and various proxies of individuals across networks and platforms, then behavioral, demographic, geo-location, and movement data must be consolidated around each individual. Now, demographic data from commercial data compilers are already standardized (one would hope), regardless of their data sources. Every other customer data type varies depending on your business.

The simplest form of transaction records would be from retail businesses, where you would sell widgets for set prices through certain channels. And what is a transaction record in that sense? “Who” bought “what,” “when,” for “how much,” through “what channel.” Even from such a simplified view point, things are not so uniform.

Let’s start with an easy one, such as common date/time stamp. Is it in form of UTC time code? That would be simple. Do we need to know the day-part of the transaction? Eventually, but by what standard? Do we need to convert them into local time of the transaction? Yes, because we need to tell evening buyers and daytime buyers apart, and we can’t use Coordinated Universal Time for that (unless you only operate in the U.K.).

“How much” isn’t so bad. It is made of net price, tax, shipping, discount, coupon redemption, and finally, total paid amount (for completed transactions). Sounds easy? Let’s just say that out of thousands of transaction files that I’ve encountered in my lifetime, I couldn’t find any “one rule” that governs how merchants would handle returns, refunds, or coupon redemptions.

Some create multiple entries for each action, with or without common transaction ID (crazy, right?). Many customer data sources contain mathematical errors all over. Inevitable file cutoff dates would create orphan records where only return transactions are found without any linkage to the original transaction record. Yes, we are not building an accounting system out of a marketing database, but no one should count canceled and returned transactions as a valid transaction for any analytics. “One-size-fits-all?” I laugh at that notion.

“Channel” may not be so bad. But at what level? What if the client has over 1,000 retail store locations all over the world? Should there be a subcategory under “Retail” as a channel? What about multiple websites with different brand names? How would we organize all that? If this type of basic – but essential – data isn’t organized properly, you won’t even be able to share store level reports with the marketing and sales teams, who wouldn’t care for a minute about “why” such basic reports are so hard to obtain.

The “what” part can be really complicated. Or, very simple if product SKUs are well-organized with proper product descriptions, and more importantly, predetermined product categories. A good sign would be the presence of a multi-level product category table, where you see entries like an apparel category broken down into Men, Women, Children, etc., and Women’s Apparel is broken down further into Formalwear, Sportswear, Casualwear, Underwear, Lingerie, Beachwear, Fashion, Accessories, etc.

For merchants with vast arrays of products, three to five levels of subcategories may be necessary even for simple BI reports, or further, advanced modeling and segmentation. But I’ve seen too many cases of incongruous and inconsistent categories (totally useless), recycled category names (really?), and weird categories such as “Summer Sales” or “Gift” (which are clearly for promotional events, not products).

All these items must be fixed and categorized properly, if they are not adequate for analytics. Otherwise, the gatekeepers of information are just dumping the hard work on poor end-users and analysts. Good luck creating any usable reports or models out of uncategorized product information. You might as well leave it as an unknown field, as product reports will have as many rows as the number of SKUs in the system. It will be a challenge finding any insights out of that kind of messy report.

Behavioral Data Are Complex and Unique to Your Business

Now, all this was about relatively simple “transaction” part. Shall we get into the online behavior data? Oh, it gets much dirtier, as any “tag” data are only as good as the person or department that tagged the web pages in question. Let’s just say I’ve seen all kinds of variations of one channel (or “Source”) called “Facebook.” Not from one place either, as they show up in “Medium” or “Device” fields. Who is going to clean up the mess?

I don’t mean to scare you, but these are just common examples in the retail industry. If you are in any subscription, continuity, travel, hospitality, or credit business, things get much more complicated.

For example, there isn’t any one “transaction date” in the travel industry. There would be Reservation Date, Booking Confirmation Date, Payment Date, Travel Date, Travel Duration, Cancellation Date, Modification Date, etc., and all these dates matter if you want to figure out what the traveler is about. If you get all these down properly and calculate distances from one another, you may be able to tell if the individual is traveling for business or for leisure. But only if all these data are in usable forms.

Always Consider Exception Management Skills

Some of you may be in businesses where turn-key solutions may be sufficient. And there are plenty of companies that provide automated, but simpler and cheaper options. The proper way to evaluate your situation would be to start with specific objectives and prioritize them. What are the functionalities you can’t live without, and what is the main goal of the data project? (Hopefully not hoarding the customer data.)

Once you set the organizational goals, try not to deviate from them so casually in the name of cost savings and automation. Your bosses and colleagues (i.e., mostly the “bottom line” folks) may not care much about the limitations of toolsets and technologies (i.e., geeky concerns).

Omnichannel marketing that requires a CDP is already complicated. So, beware of sales pitches like “All your dreams will come true with our CDP solution!” Ask some hard questions, and see if they balk at the word “customization.” Your success may depend on their ability to handle exceptions than executing some commoditized functions that they had acquired a long time ago. Unless you really believe that you will safely get to your destination on a “autopilot” mode.

 

Understanding What a Customer Data Platform Needs to Be

Marketers try to achieve holistic personalization through all conceivable channels in order to stand out among countless messages hitting targeted individuals every day, if not every hour. If the message is not clearly about the target recipient, it will be quickly dismissed. So, how can marketers achieve such an advanced level of personalization?

Modern-day marketers try to achieve holistic personalization through all conceivable channels in order to stand out among countless marketing messages hitting targeted individuals every day, if not every hour. If the message is not clearly about the target recipient, it will be quickly dismissed.

So, how can marketers achieve such an advanced level of personalization? First, we have to figure out who each target individual is, which requires data collection: What they clicked, rejected, browsed, purchased, returned, repeated, recommended, look like, complained about, etc.  Pretty much every breath they take, every move they make (without being creepy). Let’s say that you achieved that level of data collection. Will it be enough?

Enter “Customer-360,” or “360-degree View of a Customer,” or “Customer-Centric Portrait,” or “Single View of a Customer.” You get the idea. Collected data must be consolidated around each individual to get a glimpse — never the whole picture — of who the targeted individual is.

You may say, “That’s cool, we just procured technology (or a vendor) that does all that.” Considering there is no CRM database or CDP (Customer Data Platform) company that does not say one of the terms I listed above, buyers of technology often buy into the marketing pitch.

Unfortunately,the 360-degree view of a customer is just a good start in this game, and a prerequisite. Not the end goal of any marketing effort. The goal of any data project should never be just putting all available data in one place. It must support great many complex and laborious functions during the course of planning, analysis, modeling, targeting, messaging, campaigning, and attribution.

So, for the interest of marketers, allow me to share the essentials of what a CDP needs to be and do, and what the common elements of useful marketing databases are.

A CDP Must Cover Omnichannel Sources

By definition, a CDP must support all touchpoints in an omnichannel marketing environment. No modern consumer lingers around just in one channel. The holistic view cannot be achieved by just looking at their past transaction history, either (even though the past purchase behavior still remains the most powerful predictor of future behavior).

Nor do marketers have time to wait until someone buys something through a particular channel for them to take actions. All movements and indicators — as much as possible — through every conceivable channel should be included in a CDP.

Yes, some data evaporates faster than others — such as browsing history — but we are talking about a game of inches here.  Besides, data atrophy can be delayed with proper use of modeling techniques.

Beware of vendors who want to stay in their comfort zone in terms of channels. No buyer is just an online or an offline person.

Data Must Be Connected on an Individual Level

Since buyers go through all kinds of online and offline channels during the course of their journey, collected data must be stitched together to reveal their true nature. Unfortunately, in this channel-centric world, characteristics of collected data are vastly different depending on sources.

Privacy concerns and regulations regarding Personally Identifiable Information (PII) greatly vary among channels. Even if PII is allowed to be collected, there may not be any common match key, such as address, email, phone number, cookie ID, device ID, etc.

There are third-party vendors who specialize in such data weaving work. But remember that no vendor is good with all types of data. You may have to procure different techniques depending on available channel data. I’ve seen cases where great technology companies that specialized in online data were clueless about “soft-match” techniques used by direct marketers for ages.

Remember, without accurate and consistent individual ID system, one cannot even start building a true Customer-360 view.

Data Must Be Clean and Reliable

You may think that I am stating the obvious, but you must assume that most data sources are dirty. There is no pristine dataset without a serious amount of data refinement work. And when I say dirty, I mean that databases are filled with inaccurate, inconsistent, uncategorized, and unstructured data. To be useful, data must be properly corrected, purged, standardized, and categorized.

Even simple time-stamps could be immensely inconsistent. What are date-time formats, and what time zones are they in?  Dollars aren’t just dollars either. What are net price, tax, shipping, discount, coupon, and paid amounts? No, the breakdown doesn’t have to be as precise as for an accounting system, but how would you identify habitual discount seekers without dissecting the data up front?

When it comes to free-form data, things get even more complicated. Let’s just say that most non-numeric data are not that useful without proper categorization, through strict rules along with text mining. And such work should all be done up front. If you don’t, you are simply deferring more tedious work to poor analysts, or worse, to the end-users.

Beware of vendors who think that loading the raw data onto some table is good enough. It never is, unless the goal is to hoard data.

Data Must Be Up-to-Date

“Real-time update” is one of the most abused word in this business. And I don’t casually recommend it, unless decisions must be made in real-time. Why? Because, generally speaking, more frequent updates mean higher maintenance cost.

Nevertheless, real-time update is a must, if we are getting into fully automated real-time personalization. It is entirely possible to rely on trigger data for reactive personalization outside the realm of CDP environment,  but such patch work will lead to regrets most of the time. For one, how would you figure out what elements really worked?

Even if a database is not updated in real-time, most source data must remain as fresh as they can be. For instance, it is generally not recommended to append third-party demographic data real-time (except for “hot-line” data, of course). But that doesn’t mean that you can just use old data indefinitely.

When it comes to behavioral data, time really is of an essence. Click data must be updated at least daily, if not real-time.  Transaction data may be updated weekly, but don’t go over a month without updating the base, as even simple measurements like “Days since last purchase” can be way off. You all know the importance of good old recency factor in any metrics.

Data Must Be Analytics-Ready

Just because the data in question are clean and error-free, that doesn’t mean that they are ready for advanced analytics. Data must be carefully summarized onto an individual level, in order to convert “event level information” into “descriptors of individuals.”  Presence of summary variables is a good indicator of true Customer-360.

You may have all the click, view, and conversion data, but those are all descriptors of events, not people. For personalization, you need know individual level affinities (you may call them “personas”). For planning and messaging, you may need to group target individuals into segments or cohorts. All those analytics run much faster and more effectively with analytics-ready data.

If not, even simple modeling or clustering work may take a very long time, even with a decent data platform in place. It is routinely quoted that over 80% of analysts’ time go into data preparation work — how about cutting that down to zero?

Most modern toolsets come with some analytics functions, such as KPI dashboards, basic queries, and even segmentation and modeling. However, for advanced level targeting and messaging, built-in tools may not be enough. You must ask how the system would support professional statisticians with data extraction, sampling, and scoring (on the backend). Don’t forget that most analytics work fails before or after the modeling steps. And when any meltdown happens, do not habitually blame the analysts, but dig deeper into the CDP ecosystem.

Also, remember that even automated modeling tools work much better with refined data on a proper level (i.e., Individual level data for individual level modeling).

CDP Must Be Campaign-Ready

For campaign execution, selected data may have to leave the CDP environment. Sometimes data may end up in a totally different system. A CDP must never be the bottleneck in data extraction and exchange. But in many cases, it is.

Beware of technology providers that only allow built-in campaign toolsets for campaign execution. You never know what new channels or technologies will spring up in the future. While at it, check how many different data exchange protocols are supported. Data going out is as important as data coming in.

CDP Must Support Omnichannel Attribution

Speaking of data coming in and out, CDPs must be able to collect campaign result data seamlessly, from all employed channels.  The very definition of “closed-loop” marketing is that we must continuously learn from past endeavors and improve effectiveness of targeting, messaging, and channel usage.

Omnichannel attribution is simply not possible without data coming from all marketing channels. And if you do not finish the backend analyses and attribution, how would you know what really worked?

The sad reality is that a great majority of marketers fly blind, even with a so-called CDP of their own. If I may be harsh here, you are not a database marketer if you are not measuring the results properly. A CDP must make complex backend reporting and attribution easier, not harder.

Final Thoughts

For a database system to be called a CDP, it must satisfy most — if not all — of these requirements. It may be daunting for some to read through this, but doing your homework in advance will make it easier for you in the long run.

And one last thing: Do not work with any technology providers that are stingy about custom modifications. Your business is unique, and you will have to tweak some features to satisfy your unique needs. I call that the “last-mile” service. Most data projects that are labeled as failures ended up there due to a lack of custom fitting.

Conversely, what we call “good” service providers are the ones who are really good at that last-mile service. Unless you are comfortable with one-size-fits-all pre-made — but cheaper — toolset, always insist on customizable solutions.

You didn’t think that this whole omnichannel marketing was that simple, did you?

 

Don’t Blame Personalization After Messing It Up

In late 2019, Gartner predicted “80% of marketers who have invested in personalization efforts will abandon them by 2025 because of lack of ROI, the peril of customer data, or both.” But before giving up because the first few rounds didn’t pay off, shouldn’t marketers stop and think about what could have gone wrong?

In late 2019, Gartner predicted “80% of marketers who have invested in personalization efforts will abandon them by 2025 because of lack of ROI, the peril of customer data, or both.” Interesting that I started my last article quoting only about 20% of analytics works are properly applied to businesses. What is this, some 80/20 hell for marketers?

Nonetheless, the stat that I shared here begs for further questioning, especially the ROI part. Why do so many marketers think that ROI isn’t there? Simply, ROI doesn’t look good when:

  1. You invested too much money (the denominator of the ROI equation), and
  2. The investment didn’t pay off (the numerator of the same).

Many companies must have spent large sums of money on teams of specialists and service providers, data platforms featuring customer 360, personalization software (on the delivery side), analytics work for developing segments and personas, third-party data, plus the maintenance cost of it all. To justify the cost, some marginal improvements here and there wouldn’t cut it.

Then, there are attribution challenges even when there are returns. Allocating credit among all the things that marketers do isn’t very simple, especially in multichannel environments. To knock CEOs and CFOs off their chairs – basically the bottom-line people, not math or data geeks – the “credited” results should look pretty darn good. Nothing succeeds like success.

After all, isn’t that why marketers jumped onto this personalization bandwagon in the first place? For some big payoff? Wasn’t it routinely quoted that, when done right, 1:1 personalization efforts could pay off 20 times over the investment?

Alas, the key phrase here was “when done right,” while most were fixated on the dollar signs. Furthermore, personalization is a team sport, and it’s a long-term game.  You will never see that 20x return just because you bought some personalization engine and turned the default setting on.

If history taught us anything, any game that could pay off so well can’t be that simple. There are lots of in-between steps that could go wrong. Too bad that yet another buzzword is about to go down as a failure, when marketers didn’t play the game right and the word was heavily abused.

But before giving it all up just because the first few rounds didn’t pay off so well, shouldn’t marketers stop and think about what could have gone so wrong with their personalization efforts?

Most Personalization Efforts Are Reactive

If you look at so-called “personalized” messages from the customer’s point of view, most of them are just annoying. You’d say, “Are they trying to annoy me personally?”

Unfortunately, successful personalization efforts of the present day is more about pushing products to customers, as in “If you bought this, you must want that too!” When you treat your customers as mere extensions of their last purchase, it doesn’t look very personal, does it?

Ok, I know that I coveted some expensive electric guitars last time I visited a site, but must I get reminded of that visit every little turn I make on the web, even “outside” the site in question?

I am the sum of many other behaviors and interests – and you have all the clues in your database – not a hollow representation of the last click or the last purchase.  In my opinion, such one-dimensional personalization efforts ruined the term.

Personalization must be about the person, not product, brands, or channels.

Personalization Tactics Are Often Done Sporadically, Not Consistently

Reactive personalization can only be done when there is a trigger, such as someone visiting a site, browsing an item for a while, putting it in a basket without checking out, clicking some link, etc. Other than the annoyance factor I’ve already mentioned, such reactive personalization is quite limited in scale. Basically, you can’t do a damn thing if there is no trigger data coming in.

The result? You end up annoying the heck out of the poor souls who left any trail – not the vast majority for sure – and leave the rest outside the personalization universe.

Now, a 1:1 marketing effort is a number’s game. If you don’t have a large base to reach, you cannot make significant differences even with a great response rate.

So, how would you get out of that “known-data-only” trap? Venture into the worlds of “unknowns,” and convert them into “high potential opportunities” using modeling techniques. We may not know for sure if a particular target is interested in purchasing high-end home electronics, but we can certainly calculate the probability of it using all the data that we have on him.

This practice alone will increase the target base from a few percentage points to 100% coverage, as model scores can be put on every record. Now you can consistently personalize messages at a much larger scale. That will certainly help with your bottom-line, as more will see your personalized messages in the first place.

But It’s Too Creepy

Privacy concerns are for real. Many consumers are scared of know-it-all marketers, on top of being annoyed by incessant bombardments of impersonal messages; yet another undesirable side effect of heavy reliance on “known” data. Because to know for sure, you have to monitor every breath they take and every move they make.

Now, there is another added bonus of sharing data in the form of model scores. Even the most aggressive users (i.e., marketers) wouldn’t act like they actually “know” the target when all they have is a probability. When the information is given to them, like “This target is 70% likely to be interested in children’s education products,” no one would come out and say “I know you are interested in children’s education products. So, buy this!”

The key in modern day marketing is a gentle nudge, not a hard sell. Build many personas – because consumers are interested in many different things – and kindly usher them to categories that they are “highly likely” to be interested in.

Too Many Initiatives Are Set on Auto-Pilot

People can smell machines from miles away. I think humans will be able to smell the coldness of a machine even when most AIs will have passed the famous Turing Test (Definition: a test of a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human).

In the present day, detecting a machine pushing particular products is even easier than detecting a call-center operator sitting in a foreign country (not that there is anything wrong about that).

On top of that, machines are only as versatile as we set them up to be. So, don’t fall for some sales pitch that a machine can automatically personalize every message utilizing all available data. You may end up with some rudimentary personalization efforts barely superior to basic collaborative filtering, mindlessly listing all related products to what the target just clicked, viewed, or purchased.

Such efforts, of course, would be better than nothing.  For some time.  But remember that the goal is to “wow” your target customers and your bosses. Do not settle for some default settings of campaign or analytics toolsets.

Important Factors Are Ignored

When most investments are sunk in platforms, engines, and toolsets, only a little are left for tweaking, maintenance, and expansion. As all businesses are unique (even in similar industries), the last mile effort for custom fitting often makes or breaks the project. At times, unfortunately, even big items such as analytics and content libraries for digital asset management get to be ignored.

Even through a state-of-the-art AI engine, refined data works better than raw data. Your personalization efforts will fail if there aren’t enough digital assets to rotate through, even with a long list of personas and segments for everyone in the database. Basically, can you show different contents for different personas at different occasions through different media?

Data, analytics, contents, and display technologies must work harmoniously for high level personalization to work.

So What Now?

It would be a real shame if marketers hastily move away from personalization efforts when sophistication level is still elementary for the most.

Maybe we need a new word to describe the effort to pamper customers with suitable products, services and offers. Regardless of what we would call it, staying relevant to your customer is not just an option anymore. Because if you don’t, your message will categorically be dismissed as yet another annoying marketing message.

 

Data Analytics Projects Only Benefit Marketers When Properly Applied

A recent report shared that only about 20% of all analytics projects work turns out to be beneficial to businesses. Such waste. Nonetheless, is that solely the fault of data scientists? After all, even effective medicine renders useless if the patient refuses to take it.

I recently read a report that only about 20% of all analytics projects work turns out to be beneficial to businesses. Such waste. Nonetheless, is that solely the fault of data scientists? After all, even effective medicine renders useless if the patient refuses to take it.

Then again, why would users reject the results of analytics work? At the risk of gross simplification, allow me to break it down into two categories: Cases where project goals do not align with the business goals, and others where good intelligence gets wasted due to lack of capability, procedure, or will to implement follow-up actions. Basically, poor planning in the beginning, and poor execution at the backend.

Results of analytics projects often get ignored if the project goal doesn’t serve the general strategy or specific needs of the business. To put it in a different way, projects stemming from the analyst’s intellectual curiosity may or may not align with business interests. Some math geek may be fascinated by the elegance of mathematical precision or complexity of solutions, but such intrigue rarely translates directly into monetization of data assets.

In business, faster and simpler answers are far more actionable and valuable. If I ask business people if they want an answer with 80% confidence level in next 2 days, or an answer with 95% certainty in 4 weeks, the great majority would choose the quicker but less-than-perfect answer. Why? Because the keyword in all this is “actionable,” not “certainty.”

Analysts who would like to maintain a distance from immediate business needs should instead pursue pure science in the world of academia (a noble cause, without a doubt). In business settings, however, we play with data only to make tangible differences, as in dollars, cents, minutes or seconds. Once such differences in philosophy are accepted and understood by all involved parties, then the real question is: What kind of answers are most needed to improve business results?

Setting Analytics Projects Up for Success

Defining the problem statement is the hardest part for many analysts. Even the ones who are well-trained often struggle with the goal setting process. Why? Because in school, the professor in charge provides the problems to solve, and students submit solutions to them.

In business, analysts must understand the intentions of decision makers (i.e., their clients), deciphering not-so-logical general statements and anecdotes. Yeah, sure, we need to attract more high-value customers, but how would we express such value via mathematical statements? What would the end result look like, and how will it be deployed to make any difference in the end?

If unchecked, many analytics projects move forward purely based on the analysts’ assumptions, or worse, procedural convenience factors. For example, if the goal of the project is to rank a customer list in the order of responsiveness to certain product offers, then to build models like that, one may employ all kinds of transactional, behavioral, response, and demographic data.

All these data types come with different strengths and weaknesses, and even different missing data ratios. In cases like this, I’ve encountered many — too many — analysts who would just omit the whole population with missing demographic data in the development universe. Sometimes such omission adds up to be over 30% of the whole. What, are we never going to reach out to those souls just because they lack some peripheral data points for them?

Good luck convincing the stakeholders who want to use the entire list for various channel promotions. “Sorry, we can provide model scores for only 70% of your valuable list,” is not going to cut it.

More than a few times, I received questions about what analysts should do when they have to reach deep into lower model groups (of response models) to meet the demand of marketers, knowing that the bottom half won’t perform well. My response would be to forget about the model — no matter how elegant it may be — and develop heuristic rules to eliminate obvious non-targets in the prospect universe. If the model gets to be used, it is almost certain that the modeler in charge will be blamed for mediocre or bad performance, anyway.

Then I firmly warn them to ask about typical campaign size “before” one starts building some fancy models. What is the point of building a response model when the emailer would blast emails as much as he wants? To prove that the analyst is well-versed in building complex response models? What difference would it ever make in the “real” world? With that energy, it would be far more prudent to build a series of personas and product affinity models to personalize messages and offers.

Supporting Analytics Results With Marketing

Now, let’s pause for a moment and think about the second major reason why the results of analytics are not utilized. Assume that the analytics team developed a series of personas and product affinity models to customize offers on a personal level. Does the marketing team have the ability to display different offers to different targets? Via email, websites, and/or print media? In other words, do they have capabilities and resources to show “a picture of two wine glasses filled with attractive looking red wine” to people who scored high scores in the “Wine Enthusiast” model?

I’ve encountered too many situations where marketers look concerned — rather than getting excited — when talking about personas for personalization. Not because they care about what analysts must go through to produce a series of models, but because they lack creative assets and technical capabilities to make it all happen.

They often complain about lack of budget to develop multiple versions of creatives, lack of proper digital asset management tools, lack of campaign management tools that allows complex versioning, lack of ability to serve dynamic contents on websites, etc. There is no shortage of reasons why something “cannot” be done.

But, even in a situation like that, it is not the job of a data scientist to suggest increasing investments in various areas, especially when “other” departments have to cough up the money. No one gets to command unlimited resources, and every department has its own priorities. What analytics professionals must do is to figure out all kinds of limitations beyond the little world of analytics, and prioritize the work in terms of actionability.

Consider what can be done with minimal changes in the marketing ecosystem, and for preservation of analytics and marketing departments, what efforts will immediately bring tangible results? Basically, what will we be able to brag about in front of CEOs and CFOs?

When to Put Analytics Projects First

Prioritization of analytics projects should never be done solely based on data availability, ease of data crunching or modeling, or “geek” factors. It should be done in terms of potential value of the result, immediate actionability, and most importantly, alignment with overall business objectives.

The fact that only about 20% of analytics work yields business value means that 80% of the work was never even necessary. Sure, data geeks deserve to have some fun once in a while, but the fun factor doesn’t pay for the systems, toolsets, data maintenance, and salaries.

Without proper problem statements on the front-end and follow-up actions on the back-end, no amount of analytical activities would produce any value for businesses. That is why data and analytics professionals must act as translators between the business world and the technical world. Without that critical consulting layer, it becomes the-luck-of-the-draw when prioritizing projects.

To stay on target, always start with a proper analytics roadmap covering from ideation to applications stages. To be valued and appreciated, data scientists must act as business consultants, as well.

 

‘Too Much’ Is a Relative Term for Promotional Marketing

If a marketer sends you 20 promotional emails in a month, is that too much? You may say “yes” without even thinking about it. Then why did you not opt out of Amazon email programs when they send far more promotional stuff to you every month?

If a marketer sends you 20 promotional emails in a month, is that too much? You may say “yes” without even thinking about it. Then why did you not opt out of Amazon email programs when they send far more promotional stuff to you every month? Just because it’s a huge brand? I bet it’s because “some” of its promotions are indeed relevant to your needs.

Marketers are often obsessed with KPIs, such as email delivery, open, and clickthrough rates. Some companies reward their employees based on the sheer number of successful email campaign deployments and deliveries. Inevitably, such a practice leads to “over-promotions.” But does every recipient see it that way?

If a customer responds (opens, clicks, or converts, where the conversion is king) multiple times to those 20 emails, maybe that particular customer is NOT over-promoted. Maybe it is okay for you to send more promotional stuff to that customer, granted that the offers are relevant and beneficial to her. But not if she doesn’t open a single email for some time, that’s the very definition of “over-promotion,” leading to an opt-out.

As you can see, the sheer number of emails (or any other channel promotion) to a person should not be the sole barometer. Every customer is different, and recognition of such differences is the first step toward proper personalization. In other words, before worrying about customizing offers and products for a target individual, figure out her personal threshold for over-promotion. How much is too much for everyone?

Figuring out the magic number for each customer is a daunting task, so start with three basic tiers:

  1. Over-promoted,
  2. Adequately promoted, and
  3. Under-promoted.

To get to that, you must merge promotional history data (not just for emails, but for every channel) and response history data (which includes open, clickthrough, browse, and conversion data) on an individual level.

Sounds simple? But marketing organizations rarely get into such practices. Most attributions are done on a channel level, and many do not even have all required data in the same pool. Worse, many don’t have any proper match keys and rules that govern necessary matching steps (i.e., individual-level attribution).

The issue is further compounded by inconsistent rules and data availability among channels (e.g., totally different practices for online and offline channels). So much for the coveted “360-Degree Customer View.” Most organizations fail at “hello” when it comes to marrying promotion and response history data, even for the most recent month.

But is it really that difficult of an operation? After all, any respectful direct marketers are accustomed to good old “match-back” routines, complete with resolutions for fractional allocations. For instance, if the target received multiple promotions in the given study period, which one should be attributed to the conversion? The last one? The first one? Or some credit distribution, based on allocation rules? This is where the rule book comes in.

Now, all online marketers are familiar with reporting tools provided by reputable players, like Google or Adobe. Yes, it is relatively simple to navigate through them. But if the goal is to determine who is over-promoted or adequately promoted, how would you go about it? The best way, of course, is to do the match-back on an individual level, like the old days of direct marketing. But thanks to the sheer volume of online activity data and complexity of match-back, due to the frequent nature of online promotions, you’d be lucky if you could just get past basic “last-click” attribution on an individual level for merely the last quarter.

I sympathize with all of the dilemmas associated with individual-level attributions, so allow me to introduce a simpler way (i.e., a cheat) to get to the individual-level statistics of over- and under-promotion.

Step 1: Count the Basic Elements

Set up the study period of one or two years, and make sure to include full calendar years (such as rolling 12 months, 24 months, etc.). You don’t want to skew the figures by introducing the seasonality factor. Then add up all of the conversions (or transactions) for each individual. While at it, count the opens and clicks, if you have extracted data from toolsets. On the promotional side, count the number of emails and direct mails to each individual. You only have to worry about the outbound channels, as the goal is to curb promotional frequency in the end.

Step 2: Once You Have These Basic Figures, Divide ‘Number of Conversions’ by ‘Number of Promotions’

Perform separate calculations for each channel. For now, don’t worry about the overlaps among channels (i.e., double credit of conversions among channels). We are only looking for directional guidelines for each individual, not comprehensive channel attribution, at this point. For example, email responsiveness would be expressed as “Number of Conversions” divided by “Number of Email Promotions” for each individual in the given study period.

Step 3: Now That You Have Basic ‘Response Rates’

These response rates are for each channel and you must group them into good, bad, and ugly categories.

Examine the distribution curve of response rates, and break them into three segments of one.

  1. Under-promoted (the top part, in terms of response rate),
  2. Adequately Promoted (middle part of the curve),
  3. Over-promote (the bottom part, in terms of response rate).

Consult with a statistician, but when in hurry, start with one standard deviation (or one Z-score) from the top and the bottom. If the distribution is in a classic bell-curve shape (in many cases, it may not be), that will give roughly 17% each for over- and under-promoted segments, and conservatively leave about 2/3 of the target population in the middle. But of course, you can be more aggressive with cutoff lines, and one size will not fit all cases.

In any case, if you keep updating these figures at least once a month, they will automatically be adjusted, based on new data. In other words, if a customer stops responding to your promotions, she will consequently move toward the lower segments (in terms of responsiveness) without any manual intervention.

Putting It All Together

Now you have at least three basic segments grouped by their responsiveness to channel promotions. So, how would you use it?

Start with the “Over-promoted” group, and please decrease the promotional volume for them immediately. You are basically training them to ignore your messages by pushing them too far.

For the “Adequately Promoted” segment, start doing some personalization, in terms of products and offers, to increase response and value. Status quo doesn’t mean that you just repeat what you have been doing all along.

For “Under-promoted” customers, show some care. That does NOT mean you just increase the mail volume to them. They look under-promoted because they are repeat customers. Treat them with special offers and exclusive invitations. Do not ever take them for granted just because they tolerated bombardments of promotions from you. Figure out what “they” are about, and constantly pamper them.

Find Your Strategy

Why do I bother to share this much detail? Because as a consumer, I am so sick of mindless over-promotions. I wouldn’t even ask for sophisticated personalization from every marketer. Let’s start with doing away with carpet bombing to all. That begins with figuring out who is being over-promoted.

And by the way, if you are sending two emails a day to everyone, don’t bother with any of this data work. “Everyone” in your database is pretty much over-promoted. So please curb your enthusiasm, and give them a break.

Sometimes less is more.

Data Will Lead Marketers Into a New World in 2020

What will be so different in this ever-changing world, and how can marketers better prepare ourselves for the new world? Haven’t we been using data for multichannel marketing for a few decades already?

The year 2020 sounds like some futuristic time period in a science fiction novel. At the dawn of this funny sounding year, maybe it’s good time to think about where all these data and technologies will lead us. If not for the entire human collective in this short article, but at the minimum, for us marketers.

What will be so different in this ever-changing world, and how can marketers better prepare ourselves for the new world? Haven’t we been using data for multichannel marketing for a few decades already?

Every Channel Is, or Will Be Interactive 

Multichannel marketing is not a new concept, and many have been saying that every channel will become interactive medium. Then I wonder why many marketers are still acting like every channel is just another broadcasting medium for “them.” Do you really believe that marketers are still in control? That marketers can just push their agenda, the same old ways, through every channel? Uniformly? “Yeah! We are putting out this new product, so come and see!” That is so last century.

For instance, an app is not more real estate where you just hang your banners and wait for someone to click. By definition, a mobile app is an interactive medium, where information goes back and forth. And that changes the nature of the communication from “We talk, they listen” to “We listen first, and then we talk based on what we just heard.”

Traditional media will go through similar changes. Even the billboards on streets, in the future, will be customized based on who’s seeing it. Young people don’t watch TV in the old-fashioned way, mindlessly flipping through channels like their parents. They will actively seek out content that suites “them,” not the other way around. And in such an interactive world, the consumers of the content have all the power. They will mercilessly stop, cut out, opt out, and reject anything that is even remotely boring to “them.”

Marketers are not in charge of communication anymore. They say an average human being looks at six to seven different screens every day. And with wearable devices and advancement in mobile technologies, even the dashboard on a car will stop being just a dumb dashboard. What should marketers do then? Just create another marketing department called “wearable division,” like they created the “email marketing” division?

The sooner marketers realize that they are not in charge, but the consumers are, the better off they would be. Because with that realization, they will cease to conduct channel marketing the way they used to do, with extremely channel-centric mindsets.

When the consumers are in charge, we must think differently. Everything must be customer-centric, not channel- or division-centric. Know that we can be cut off from any customer anytime through any channel, if we are more about us than about them.

Every Interaction Will Be Data-based, and in Real-time

Interactive media leave ample amounts of data behind every interaction. How do you think this word “Big Data” came about? Every breath we take and every move we make turn into piles of data somewhere. That much is not new.

What is new is that our ability to process and dissect such ample amounts of data is getting better and faster, at an alarming rate. So fast that we don’t even say words like Big Data anymore.

In this interactive world, marketers must listen first, and then react. That listening part is what we casually call data-mining, done by humans and machines, alike. Without ploughing through data, how will we even know what the conversation is about?

Then the second keyword in the subheading is “real-time.” Not only do we have to read our customers’ behavior through breadcrumbs they leave behind (i.e., their behavioral data), we must do it incredibly fast, so that our responses seem spontaneous. As in “Oh, you’re looking for a set of new noise-canceling earbuds! Here are the ones that you should consider,” all in real-time.

Remember the rule No. 1 that customers can cut us out anytime. We may have less than a second before they move on.

Marketers Must Stay Relevant to Cut Through the Noise

Consumers are bored to tears with almost all marketing messages. There are too many of them, and most aren’t about the readers, but the pushers. Again, it should be all about the consumers, not the sellers.

It stops being entirely boring when the message is about them though. Everybody is all about themselves, really. If you receive a group photo that includes you, whose face would you check out first? Of course, your own, as in “Hmm, let me see how I look here.”

That is the fundamental reason why personalization works. But only if it’s done right.

Consumers can smell fake intimacy from miles away. Young people are particularly good at that. They think that the grownups don’t understand social media at all for that reason. They just hate it when someone crashes a party to hard-sell something. Personalization is about knowing your targets’ affinities and suggesting — not pushing — something that may suite “them.” A gentle nudge, but not a hard sell.

With ample amounts of data all around, it may be very tempting to show how much we know about the customers. But never cross that line of creepiness. Marketers must be relevant to stay connected, but not overly so. It is a fine balance that we must maintain to not be ignored or rejected.

Machine Learning and AI Will Lead to Automation on All Fronts

To stay relevant at all times, using all of the data that we have is a lot of work. Tasks that used to take months — from data collection and refinement to model-based targeting and messaging — should be done in minutes, if not seconds. Such a feat isn’t possible without automation. On that front, things that were not imaginable only a few years ago are possible through advancement in machine learning or AI, in general.

One important note for marketers who may not necessarily be machine learning specialists is that what the machines are supposed to do is still up to the marketers, not the machines. Always set the goals first, have a few practice rounds in more conventional ways, and then get on a full automation mode. Otherwise, you may end up automating wrong practices. You definitely don’t want that. And, more importantly, target consumers would hate that. Remember, they hate fake intimacy, and more so if they smell cold algorithms in play along the way.

Huge Difference Between Advanced Users and Those Who Are Falling Behind

In the past, many marketers considered data and analytics as optional items, as in “Sure, they sound interesting, and we’ll get around to it when we have more time to think about it.” Such attitudes may put you out of business, when giants like Amazon are eating up the world with every bit of computing power they have (not that they do personalization in an exemplary way all of the time).

If you have lines of products that consumers line up to buy, well, all the more power to you. And, by all means, don’t worry about pampering them proactively with data. But if you don’t see lines around the block, you are in a business that needs to attract new customers and retain existing customers more effectively. And such work is not something that you can just catch up on in a few months. So get your data and targeting strategy set up right away. I don’t believe in new year’s resolutions, but this month being January and all, you might as well call it that.

Are You Ready for the New World?

In the end, it is all about your target customers, not you. Through data, you have all the ammunition that you need to understand them and pamper them accordingly. In this age, marketers must stay relevant with their targets through proper personalization at all stages of the customer journey. It may sound daunting, but all of the technologies and techniques are ripe for such advanced personalization. It really is about your commitment — not anything else.

Marketers Find the Least-Wrong Answers Via Modeling

Why do marketers still build models when we have ample amounts of data everywhere? Because we will never have every piece of data about everything. We just don’t know what we don’t know.

Why do marketers still build models when we have ample amounts of data everywhere? Because we will never have every piece of data about everything. We just don’t know what we don’t know.

Okay, then — we don’t get to know about everything, but what are the data that we possess telling us?

We build models to answer that question. Even scientists who wonder about the mysteries of the universe and multiverses use models for their research.

I have been emphasizing the importance of modeling in marketing through this column for a long time. If I may briefly summarize a few benefits here:

  • Models Fill in the Gaps, covering those annoying “unknowns.” We may not know for sure if someone has an affinity for luxury gift items, but we can say that “Yes, with data that we have, she is very likely to have such an affinity.” With a little help from the models, the “unknowns” turn into “potentials.”
  • Models Summarize Complex Data into simple-to-use “scores.” No one has time to dissect hundreds of data variables every time we make a decision. Model scores provide simple answers, such as “Someone likely to be a bargain-seeker.” Such a model may include 10 to 20 variables, but the users don’t need to worry about those details at the time of decision-making. Just find suitable offers for the targets, based on affinities and personas (which are just forms of models).
  • Models are Far More Accurate Than Human Intuition. Even smart people can’t imagine interactions among just two or three variables in their heads. Complex multivariate interaction detection is a job for a computer.
  • Models Provide Consistent Results. Human decision-makers may get lucky once in a while, but it will be hard to keep it up with machines. Mathematics do not fluctuate too much in terms of performance, provided with consistent and accurate data feeds.
  • Models Reveal Hidden Patterns in data. When faced with hundreds of data variables, humans often resort to what they are accustomed to (often fewer than four to five factors). Machines indiscriminately find new patterns, relentlessly looking for the best suitable answers.
  • Models Help Expand the Targeting Universe. If you want a broader target, just go after slightly lower score targets. You can even measure the risk factors while in such an expansion mode. That is not possible with some man-made rules.
  • When Done Right, Models Save Time and Effort. Marketing automation gets simpler, too, as even machines can tell high and low scores apart easily. But the keywords here are “when done right.”

There are many benefits of modeling, even in the age of abundant data. The goal of any data application is to help in the decision-making process, not aid in hoarding the data and bragging about it. Do you want to get to the accurate, consistent, and simple answers — fast? Don’t fight against modeling, embrace it. Try it. And if it doesn’t work, try it in another way, as the worst model often beats man-made rules, easily.

But this time, I’m not writing this article just to promote the benefits of modeling again. Assuming that you embrace the idea already, let’s now talk about the limitations of it. With any technique, users must be fully aware of the downsides of it.

It Mimics Existing Patterns

By definition, models identify and mimic the patterns in the existing data. That means, if the environment changes drastically, all models built in the old world will be rendered useless.

For example, if there are significant changes in the supply chain in a retail business, product affinity models built for old lines of products won’t work anymore (even if products may look similar). More globally, if there were major disruptions, such as a market crash or proliferation of new technologies, none of the old assumptions would continue to be applicable.

The famous economics phrase Ceteris paribus — all other things being equal — governs conventional modeling. If you want your models to be far more adaptive, then consider total automation of modeling through machine learning. But I still suggest trying a few test models in an old-fashioned way, before getting into a full automation mode.

If the Target Is Off, Everything Is Off

If the target mark is hung on a wrong spot, no sharpshooter will be able to hit the real target. A missile without a proper guidance system is worse than not having one at all. Setting the right target for a model is the most critical and difficult part in the whole process, requiring not only technical knowledge, but also deep understanding of the business at stake, the nature of available data, and the deployment mechanism at the application stage.

This is why modeling is often called “half science, half art.” A model is only as accurate as the target definition of the model. (For further details on this complex subject, refer to “Art of Targeting”).

The Model Is Only as Good as the Input Data

No model can be saved if there are serious errors or inconsistencies in the data. It is not just about bluntly wrong data. If the nature of the data is not consistent between the model development sample and the practical pool of data (where the model will be applied and used), the model in question will be useless.

This is why the “Analytics Sandbox” is important. Such a sandbox environment is essential — not just for simplification of model development, but also for consistent application of models. Most mishaps happen before or after the model development stage, mostly due to data inconsistencies in terms of shapes and forms, and less due to sheer data errors (not that erroneous data is acceptable).

The consistency factor matters a lot: If some data variables are “consistently” off, they may still possess some predictive power. I would even go as far as stating that consistency matters more than sheer accuracy.

Accuracy Is a Relative Term

Users often forget this important fact, but model scores aren’t pinpoint accurate all of the time. Some models are sharper than others, too.

A model score is just the best estimate with the existing data. In other words, we should take model scores as the least-wrong answers in a given situation.

So, when I say it is accurate, I mean to say a model is more accurate than human intuition based on a few basic data points.

Therefore, the user must always consider the risk of being wrong. Now, being wrong about “Who is more likely to respond to this 15% discount offer?” is a lot less grave than being wrong about “Who is more likely to be diabetic?”

In fact, if I personally face such a situation, I won’t even recommend building the latter model, as the cost of being wrong is simply too high. (People are very sensitive about their medical information.) Some things should not just be estimated.

Even with innocuous models, such as product affinities and user propensities, users should never treat them as facts. Don’t act like you “know” the target, simply because some model scores are available to you. Always approach your target with a gentle nudge; as in, “I don’t know for sure if you would be interested in our new line of skin care products, but would you want to hear more about it?” Such gentle approaches always sound friendlier than acting like you “know” something about them for sure. That seems just rude on the receiving end, and recipients of blunt messages may even think that you are indeed creepy.

Users sometimes make bold moves with an illusion that data and analytics always provide the right answers. Maybe the worst fallacy in the modern age is the belief that anything a computer spits out is always correct.

Users Abuse Models

Last month, I shared seven ways users abuse models and ruin the results (refer to “Don’t Ruin Good Models by Abusing Them”). As an evangelist of modeling techniques, I always try to prevent abuse cases, but they still happen in the application stages. All good intentions of models go out the window if they are used for the wrong reasons or in the wrong settings.

I am not at all saying that anyone should back out of using models in their marketing practices for the shortfalls that I listed here. Nonetheless, to be consistently successful, users must be aware of limitations of models, as well. Especially if you are about to go on full marketing automation. With improper application of models, you may end up automating bad or wrong practices really fast. For the sake of customers on the receiving end — not just for the safety of your position in the marketing industry — please be more careful with this sharp-edged tool called modeling.